Splitting state-county-MSA string variable - r

Question

I have a variable below that I believe is separated by space.
[95] "Florida Volusia Deltona-Daytona Beach-Ormond Beach"
[96] "Iowa Polk Des Moines"
[97] "Mississippi Hinds Jackson"
[98] "Georgia Richmond Augusta-Richmond"
[99] "Ohio Mahoning Youngstown-Warren-Boardman"
[100] " Pennsylvania Lackawanna Scranton--Wilkes-Barre"
[101] " Pennsylvania Dauphin Harrisburg-Carlisle"
[102] " Florida Brevard Palm Bay-Melbourne-Titusville"
[103] " Utah Utah Provo-Orem"
[104] " Tennessee Hamilton Chattanooga"
[105] " North Carolina Durham Durham"
I want to create three variables out of this string - state, county, and MSA. But the usual string split commands are not working. I tried stringi command too but fail to split the variable. Not sure why this is happening, as the command works on simpler strings.
> strsplit(BK_state_county_MSA$non_squished_states_county_MSA_names_df,"")
Error in strsplit(BK_state_county_MSA$non_squished_states_county_MSA_names_df, :
non-character argument
> BK<-strsplit(as.character(BK_state_county_MSA$non_squished_states_county_MSA_names_df),"\\t")
> str(BK) #List of 0
list()
> stri_split(str=BK_state_county_MSA$non_squished_states_county_MSA_names_df, regex="\\t",n=3)
list()
> BK <-stri_split_lines(BK_state_county_MSA$non_squished_states_county_MSA_names_df)
list()
EDIT - The original data has 104 observations, but I am posting only 8 observations with dput command...
> dput(BK_state_county_MSA)
structure(list(non_squished_state_county_MSA = c(
"New York Bronx New York-Wayne-White Plains",
"New York Kings New York-Wayne-White Plains",
" Pennsylvania Lackawanna Scranton--Wilkes-Barre",
" Pennsylvania Dauphin Harrisburg-Carlisle",
" Florida Brevard Palm Bay-Melbourne-Titusville",
" Utah Utah Provo-Orem",
" Tennessee Hamilton Chattanooga",
" North Carolina Durham Durham")), row.names = c(NA,
-8L), class = "data.frame")

Answer 1

Here's an option using stri_trim_left() from stringi and separate() from tidyr:
stri_trim_left() removes strings with a leading whitespace, which occurs in your data starting at [100]. You can then separate() the strings into the three specified columns state, country and MSA, separated by at least 2 whitespaces (sep = " {2,}").
Data
BK_state_county_MSA<- structure(list(non_squished_state_county_MSA = c(
"New York Bronx New York-Wayne-White Plains",
"New York Kings New York-Wayne-White Plains",
" Pennsylvania Lackawanna Scranton--Wilkes-Barre",
" Pennsylvania Dauphin Harrisburg-Carlisle",
" Florida Brevard Palm Bay-Melbourne-Titusville",
" Utah Utah Provo-Orem",
" Tennessee Hamilton Chattanooga",
" North Carolina Durham Durham")), row.names = c(NA,
-8L), class = "data.frame")
Code
library(tidyr)
library(stringi)
BK_state_county_MSA %>% mutate(non_squished_state_county_MSA = stri_trim_left(non_squished_state_county_MSA)) %>%
separate(non_squished_state_county_MSA, into = c("state", "country", "MSA"), sep = " {2,}")
Output
state country MSA
1 New York Bronx New York-Wayne-White Plains
2 New York Kings New York-Wayne-White Plains
3 Pennsylvania Lackawanna Scranton--Wilkes-Barre
4 Pennsylvania Dauphin Harrisburg-Carlisle
5 Florida Brevard Palm Bay-Melbourne-Titusville
6 Utah Utah Provo-Orem
7 Tennessee Hamilton Chattanooga
8 North Carolina Durham Durham

Answer 2

There's probably something useful, in terms of further data analysis, about keeping North with Carolina and New with York & etc. And it's always nice to have a one liner, but sometimes a few lines get you where is best for moving forward. Consider playing around like this:
maddening_txt <- " North Carolina Durham Durham"
strsplit(maddening_txt, split = ' ') # n space = 4L
[[1]]
[1] " North Carolina" " Durham" "" ""
[5] "" " Durham"
nchar(strsplit(maddening_txt, split = ' ')[[1]])
[1] 15 7 0 0 0 7
# you could throw in a which test for >0 for indexing here
strsplit(maddening_txt, split = ' ')[[1]][c(1,2,6)]
[1] " North Carolina" " Durham" " Durham"
string_vec <- strsplit(maddening_txt, split = ' ')[[1]][c(1,2,6)]
> string_vec[1]
[1] " North Carolina"
string_vec[2:3]
[1] " Durham" " Durham"
Left as it is, when returned to in the future, something like split=' ' begs 'How many spaces was that?'
So, more explicitly, and using some of the above notation:
> strsplit(sc_msa$entries[8], split='\\s{5,}')
[[1]]
[1] " North Carolina" "Durham" "Durham"
This appears to also remove leading left white space without recourse to trim. And then the above mentioned
for
sc_msa_lst <- list()
> for(i in 1:length(sc_msa$entries)) {
+ sc_msa_lst[[i]] <- strsplit(sc_msa$entries[i], split='\\s{5,}')
+ }
> sc_msa_lst
[[1]]
[[1]][[1]]
[1] "New York" "Bronx"
[3] "New York-Wayne-White Plains"
[[2]]
[[2]][[1]]
[1] "New York" "Kings"
[3] "New York-Wayne-White Plains"
[[3]]
[[3]][[1]]
[1] " Pennsylvania" "Lackawanna" "Scranton--Wilkes-Barre"
[[4]]
[[4]][[1]]
[1] " Pennsylvania" "Dauphin" "Harrisburg-Carlisle"
[[5]]
[[5]][[1]]
[1] " Florida" "Brevard"
[3] "Palm Bay-Melbourne-Titusville"
[[6]]
[[6]][[1]]
[1] " Utah" "Utah" "Provo-Orem"
[[7]]
[[7]][[1]]
[1] " Tennessee" "Hamilton" "Chattanooga"
[[8]]
[[8]][[1]]
[1] " North Carolina" "Durham" "Durham"
So, our regex pattern, '\s{5,}' works for cases seen so far. Next, I have to remember how to make a data.frame...there's got to be a list of lists to data.frame question on SO somewhere that'll help me.
A completely unacceptable answer, but it will get you to the guts of what simplifications are offering and provide more control going forward.
Data, with renaming:
sc_msa <- structure(list(entries = c("New York Bronx New York-Wayne-White Plains",
"New York Kings New York-Wayne-White Plains",
" Pennsylvania Lackawanna Scranton--Wilkes-Barre",
" Pennsylvania Dauphin Harrisburg-Carlisle",
" Florida Brevard Palm Bay-Melbourne-Titusville",
" Utah Utah Provo-Orem", " Tennessee Hamilton Chattanooga",
" North Carolina Durham Durham")), row.names = c(NA,
-8L), class = "data.frame")